Abstract: The present invention provides a signal fusion system that is a probabilistic system fusing an arbitrary combination of heterogeneous signals for determining the location of electronic devices. The system includes a signal sampling device for detecting one or more signals emitted by an electronic device to be located, including geolocation signals, WiFi signals, Bluetooth signals, 4G communication signals, 5G communication signals, geomagnetism signals, or inertial navigation system signals (INS). A likelihood processor cooperates with the signal sampling device to receive information about selected sampled signals, and creates a grid of reference points for an interested area in which the electronic device may be located. The likelihood processor independently computes, for each selected sampled signal, a location likelihood that is a probability of observing the sampled signal given that the electronic device is located at different reference points in the grid.

Abstract: The present disclosure describes a semiconductor structure and a method for forming the same. The method can include forming a fin structure over a substrate. The fin structure can include a channel layer and a sacrificial layer. The method can further include forming a first recess structure in a first portion of the fin structure, forming a second recess structure in the sacrificial layer of a second portion of the fin structure, forming a dielectric layer in the first and second recess structures, and performing an oxygen-free cyclic etching process to etch the dielectric layer to expose the channel layer of the second portion of the fin structure. The oxygen-free cyclic etching process can include two etching processes to selectively etch the dielectric layer over the channel layer.

Abstract: A VSR approach with temporal consistency using generative adversarial networks (VistGAN) that requires only the training HR video sequence to generate the HR/LR video frame pairs, instead of the pre-artificial-synthesized HR/LR video frame pairs, for training. By this unsupervised learning method, the encoder degrades the input HR video frames of a training HR video sequence to their LR counterparts, and the decoder seeks to recover the original HR video frames from the LR video frames. To improve the temporal consistency the unsupervised learning method provides a sliding window that explores the temporal correlation in both HR and LR domains. It keeps the temporal consistent and also fully utilizes high-frequency details from the last-generated reconstructed HR video frame.

Abstract: The present disclosure describes a semiconductor structure and a method for forming the same. The method can include forming a fin structure over a substrate. The fin structure can include a channel layer and a sacrificial layer. The method can further include forming a first recess structure in a first portion of the fin structure, forming a second recess structure in the sacrificial layer of a second portion of the fin structure, forming a dielectric layer in the first and second recess structures, and performing an oxygen-free cyclic etching process to etch the dielectric layer to expose the channel layer of the second portion of the fin structure. The oxygen-free cyclic etching process can include two etching processes to selectively etch the dielectric layer over the channel layer.

Abstract: An architecture is provided for cooperative target tracking and signal propagation learning using mobile sensors. A method can comprise as a function of sensing data representative of a location of a target device at a first defined moment and model data relating to a motion model representing a probability density function, determining, by a system comprising a processor, a group of locations for the target device at a second defined time point, wherein the probability density function facilitates determining, based on the location of the target device at the first defined moment, a current location of the target device at a third defined moment; and as a function of the group of locations, generating, by the system, a data structure representing a matrix of received signal strength values; and identifying, by the system, a location of the group of locations for the target device at the third defined moment based on the data structure.

Abstract: Various embodiments disclosed herein enable a mobile device to accurately locate where the mobile device is located indoors by measuring geomagnetic fields, and tracing relative changes in the magnetic fields as the mobile device is moved. The relative changes can be compared to a fingerprint signal map, and a trace of a portion of the path can be determined by matching the relative changes in the magnetic fields to relative changes in the fingerprint signal map. A collection of the path portions can then be connected by determining a shortest path that connects each of the path portions. In another embodiment, a step counter can also be used to constrain possible locations, and fuse the geomagnetic field and step counter information for joint indoor localization.

Abstract: The present invention provides a signal fusion system that is a probabilistic system fusing an arbitrary combination of heterogeneous signals for determining the location of electronic devices. The system includes a signal sampling device for detecting one or more signals emitted by an electronic device to be located, including geolocation signals, WiFi signals, Bluetooth signals, 4G communication signals, 5G communication signals, geomagnetism signals, or inertial navigation system signals (INS). A likelihood processor cooperates with the signal sampling device to receive information about selected sampled signals, and creates a grid of reference points for an interested area in which the electronic device may be located. The likelihood processor independently computes, for each selected sampled signal, a location likelihood that is a probability of observing the sampled signal given that the electronic device is located at different reference points in the grid.

Abstract: The present disclosure describes a semiconductor structure and a method for forming the same. The method can include forming a fin structure over a substrate. The fin structure can include a first channel layer and a sacrificial layer. The method can further include forming a first recess structure in a first portion of the fin structure, forming a second recess structure in the sacrificial layer of a second portion of the fin structure, forming a dielectric layer in the first and second recess structures, and performing an oxygen-free cyclic etching process to etch the dielectric layer to expose the channel layer of the second portion of the fin structure.

Abstract: A VSR approach with temporal consistency using generative adversarial networks (VistGAN) that requires only the training HR video sequence to generate the HR/LR video frame pairs, instead of the pre-artificial-synthesized HR/LR video frame pairs, for training. By this unsupervised learning method, the encoder degrades the input HR video frames of a training HR video sequence to their LR counterparts, and the decoder seeks to recover the original HR video frames from the LR video frames. To improve the temporal consistency the unsupervised learning method provides a sliding window that explores the temporal correlation in both HR and LR domains. It keeps the temporal consistent and also fully utilizes high-frequency details from the last-generated reconstructed HR video frame.

Abstract: Constructing and/or recovering paths of mobile devices is enabled. For instance, a method comprises: receiving, by wireless fidelity (Wi-Fi) sensors, respective probes from mobile devices, grouping probes having a same media access control (MAC) address into segments, identifying fingerprints of information elements the mobile devices, grouping segments according to an identified fingerprint, determining an increment of a sequence number corresponding to consecutive segments of a segment group, determining a time gap between the sequence number corresponding to a first probe and an incremented sequence number corresponding to a second probe having a timestamp that is later in time than the first probe, and comparing a growth rate of the sequence number corresponding to the consecutive segments to determine a forward segment of the consecutive segments, resulting in a constructed path of the mobile devices, and storing the constructed path in a database.

Abstract: Various implementations disclosed herein include a thermal management system suitable electronic devices. In some implementations, a thermal management system includes an air guide and a cage wall, which together provide: an air intake having a first airflow area to produce a first air pressure; an airflow constriction, following the air intake, having a second airflow area smaller than the first airflow area to produce a second air pressure, and defining a first region; and an outlet following the airflow constriction having a third airflow area greater than the second airflow area to produce a third air pressure. The device cage has at least one aperture in the first region. The second air pressure is less than the first air pressure and the third air pressure and is sufficiently low to draw heated air from within the device cage through the at least one aperture to be expelled through the outlet.

Abstract: Various embodiments disclosed herein enable a mobile device to accurately locate where the mobile device is located indoors by measuring geomagnetic fields, and tracing relative changes in the magnetic fields as the mobile device is moved. The relative changes can be compared to a fingerprint signal map, and a trace of a portion of the path can be determined by matching the relative changes in the magnetic fields to relative changes in the fingerprint signal map. A collection of the path portions can then be connected by determining a shortest path that connects each of the path portions. In another embodiment, a step counter can also be used to constrain possible locations, and fuse the geomagnetic field and step counter information for joint indoor localization.

Abstract: An architecture is provided for cooperative target tracking and signal propagation learning using mobile sensors. A method can comprise as a function of sensing data representative of a location of a target device at a first defined moment and model data relating to a motion model representing a probability density function, determining, by a system comprising a processor, a group of locations for the target device at a second defined time point, wherein the probability density function facilitates determining, based on the location of the target device at the first defined moment, a current location of the target device at a third defined moment; and as a function of the group of locations, generating, by the system, a data structure representing a matrix of received signal strength values; and identifying, by the system, a location of the group of locations for the target device at the third defined moment based on the data structure.

Abstract: Constructing and/or recovering paths of mobile devices is enabled. For instance, a method comprises: receiving, by wireless fidelity (Wi-Fi) sensors, respective probes from mobile devices, grouping probes having a same media access control (MAC) address into segments, identifying fingerprints of information elements the mobile devices, grouping segments according to an identified fingerprint, determining an increment of a sequence number corresponding to consecutive segments of a segment group, determining a time gap between the sequence number corresponding to a first probe and an incremented sequence number corresponding to a second probe having a timestamp that is later in time than the first probe, and comparing a growth rate of the sequence number corresponding to the consecutive segments to determine a forward segment of the consecutive segments, resulting in a constructed path of the mobile devices, and storing the constructed path in a database.

Abstract: When access point signals are altered, conventional fingerprint-based indoor localization techniques are note accuracy. Localization with altered access point and fingerprint updating can achieves accurate indoor localization and automatically update a fingerprint database with altered access points. Using subset sampling, the system detect the altered access points, filter them out by a received signal strength vector and find the location of a client. Given the received signal strength vectors received and the estimated location, the system can update a fingerprint database with the signal changes by applying a non-parametric Gaussian process regression method.

Abstract: Various implementations disclosed herein include a thermal management system suitable electronic devices. In some implementations, a thermal management system includes an air guide and a cage wall, which together provide: an air intake having a first airflow area to produce a first air pressure; an airflow constriction, following the air intake, having a second airflow area smaller than the first airflow area to produce a second air pressure, and defining a first region; and an outlet following the airflow constriction having a third airflow area greater than the second airflow area to produce a third air pressure. The device cage has at least one aperture in the first region. The second air pressure is less than the first air pressure and the third air pressure and is sufficiently low to draw heated air from within the device cage through the at least one aperture to be expelled through the outlet.

Abstract: Various implementations disclosed herein include a thermal management system suitable electronic devices. In some implementations, a thermal management system includes an air guide and a cage wall, which together provide: an air intake having a first airflow area to produce a first air pressure; an airflow constriction, following the air intake, having a second airflow area smaller than the first airflow area to produce a second air pressure, and defining a first region; and an outlet following the airflow constriction having a third airflow area greater than the second airflow area to produce a third air pressure. The device cage has at least one aperture in the first region. The second air pressure is less than the first air pressure and the third air pressure and is sufficiently low to draw heated air from within the device cage through the at least one aperture to be expelled through the outlet.

Abstract: When access point signals are altered, conventional fingerprint-based indoor localization techniques are note accuracy. Localization with altered access point and fingerprint updating can achieves accurate indoor localization and automatically update a fingerprint database with altered access points. Using subset sampling, the system detect the altered access points, filter them out by a received signal strength vector and find the location of a client. Given the received signal strength vectors received and the estimated location, the system can update a fingerprint database with the signal changes by applying a non-parametric Gaussian process regression method.

Abstract: Locating mobile devices based on a junction of signal tiles is disclosed herein. Mobile devices are mapped to a convex hull within which the mobile device is likely to be located in based on a received signal strength indicator (RSSI) associated with an access point (AP). Using the RSSI from a set of APs, the mobile device can be located within a junction of a set of convex hulls, or tiles using a linear programming function. To mitigate signal noise and improve the accuracy of the location determination, the localization system can use the RSSIs of the APs that have signals with a predetermined signal dynamic range that leads to smaller tiles. Additionally, the localization system can reduce the search complexity of matching fingerprints to APs by partitioning the space into clusters with reference points that have similar signal vectors.

Abstract: Video on demand is orchestrated between systems by linear program (LP) based multimedia storage and retrieval. An LP based storage and retrieval system can partition multimedia data into an integral number of segments. Further, the LP based storage and retrieval system can store a first amount of the integral number of segments in a storage device of storage devices of a storage network in response to a determination that a storage size of the storage device satisfies a first condition with respect to the first amount of the integral number of segments. Furthermore, such system can store a segment of a second amount of the integral number of segments in a remote storage device of the remote storage devices in response to a determination that the segment satisfies a second condition with respect to a defined number of remote storage devices of the storage devices that have stored the segment.